Scheduling & synchronisation

[Mandar Mitra]

QUESTION : One possible correction that I thought of while going through the notes. In the first solution of n process synchronization problem (using test&set lock and want[n] ), you had mentioned that bounded waiting is contingent on the fact that the scheduler itself is starvation free. But I think that bounded waiting should be guaranteed regardless of the scheduler, only progress should be dependent on whether the scheduler is starvation free or not.

ANS:

Yes, excellent point! I suppose what you're thinking is: even if the scheduler is not fair, the other processes don't get to enter the CS more than once before the process in question. That is not to say that 'our' process gets to enter the CS within a bounded amount of time. For that, progress is necessary, for which fair scheduling is necessary. Is this understanding correct? 

[Sreyan Saha]

Yes this train of thought is indeed what led me to that conclusion. Is my understanding correct regarding this ?